fix: improved clonality event definition, usage of major subclone for…

… assembly polishing; upgrade to varlociraptor 6.0 (#585)

* fix: improved clonality event definition, usage of major subclone for assembly polishing; upgrade to varlociraptor 6.0

* adapt wildcard constraints

workflow/envs/varlociraptor.yaml

@@ -3,4 +3,4 @@ channels:
   - bioconda
   - nodefaults
 dependencies:
-  - varlociraptor =5.3
+  - varlociraptor =6.0

workflow/rules/assembly.smk

@@ -159,8 +159,8 @@ rule filter_chr0:
 rule assembly_polishing_illumina:
     input:
         fasta="results/{date}/contigs/ordered/{sample}.fasta",
-        bcf="results/{date}/filtered-calls/ref~{sample}/{sample}.clonal.nofilter.orf.bcf",
-        bcfidx="results/{date}/filtered-calls/ref~{sample}/{sample}.clonal.nofilter.orf.bcf.csi",
+        bcf="results/{date}/filtered-calls/ref~{sample}/{sample}.subclonal-major.nofilter.orf.bcf",
+        bcfidx="results/{date}/filtered-calls/ref~{sample}/{sample}.subclonal-major.nofilter.orf.bcf.csi",
     output:
         report(
             "results/{date}/polishing/bcftools-illumina/{sample}.fasta",

workflow/rules/common.smk

@@ -599,12 +599,16 @@ def get_bwa_index(wildcards):
 
 
 def get_target_events(wildcards):
-    if wildcards.reference == "main" or wildcards.clonality != "clonal":
-        # calling variants against the wuhan reference or we are explicitly interested in subclonal as well
-        return "SUBCLONAL_MINOR SUBCLONAL_MAJOR SUBCLONAL_HIGH CLONAL"
-    else:
-        # only keep clonal variants
+    if wildcards.clonality == "clonal":
         return "CLONAL"
+    elif wildcards.clonality == "subclonal-major":
+        return "CLONAL SUBCLONAL_MAJOR SUBCLONAL_HIGH"
+    elif wildcards.clonality == "subclonal-high":
+        return "CLONAL SUBCLONAL_HIGH"
+    elif wildcards.clonality == "subclonal":
+        return "CLONAL SUBCLONAL_MAJOR SUBCLONAL_HIGH SUBCLONAL_MINOR"
+    else:
+        raise ValueError(f"Unsupported clonality value: {wildcards.clonality}")
 
 
 def get_control_fdr_input(wildcards):
@@ -1670,7 +1674,7 @@ def get_genome_annotation(suffix=""):
 wildcard_constraints:
     sample="[^/.]+",
     vartype="|".join(VARTYPES),
-    clonality="subclonal|clonal",
+    clonality="subclonal|clonal|subclonal-major|subclonal-high",
     annotation="orf|protein",
     filter="|".join(
         list(map(re.escape, config["variant-calling"]["filters"])) + ["nofilter"]

workflow/rules/variant_filtration.smk

@@ -35,7 +35,7 @@ rule control_fdr:
     conda:
         "../envs/varlociraptor.yaml"
     shell:
-        "varlociraptor filter-calls control-fdr --local {input} --var {wildcards.vartype} "
+        "varlociraptor filter-calls control-fdr --mode local-smart {input} --var {wildcards.vartype} "
         "--events {params.events} --fdr {params.fdr} > {output} 2> {log}"